nature biotechnology • VOLUME 20 • SEPTEMBER 2002 • www.nature.com/naturebiotechnology

NEWS AND VIEWS

884

tioned carboxyl groups. This imprinted den-drimer was shown to selectively rebind struc-tural analogs of the template, for example theisomeric compound tetrakis-meso(2,6-dihy-droxyphenyl)-porphyrine, although for spa-tial reasons not the template itself.

Have Zimmermann and colleagues madean imprinter’s dream come true? At firstglance, the answer seems to be yes: theresearchers produced a low-molecular-massMIP (∼ 104 Da), that is soluble in commonorganic solvents and has only one well-defined, and apparently readily accessible,binding site for the template per MIP mole-cule. The team postulates that binding of thetarget by the MIP involves a dynamic “breath-ing process,” indicating that the MIP has someflexibility in its structure. Indeed, accessibilityis an advantage of these imprinted den-drimers, allowing an easy and completeremoval of the template that is not always pos-sible with traditional MIPs because of theirhighly crosslinked and compact structure.

First employed by Wulff2, covalentlyimprinted complexes are more stable andstructurally better defined than noncovalentMIPs. The resulting material therefore con-tains a population of binding sites with a moreor less homogeneous affinity for their targets.Zimmermann and colleagues took covalentlygenerated MIPs one step further—producingMIPs with just one binding site per molecule.The originality of their system lies in how theygenerated the imprinted molecules: they pre-synthesized all parts of the molecules, assem-bled them with the help of the template, andstabilized the whole by crosslinking the outershell. Although they relied on covalent bond-ing during the imprinting process, rebindingof the target molecule by the MIP occursthrough noncovalent interactions.Whitcombe and colleagues used a similarstrategy to imprint a peptide using monomerscovalently linked to the target molecule viashort sacrificial spacers3. This creates space in

Like a hand in a glove, specialized structuressuch as antibodies, hormone receptors, andenzymes fit perfectly with their natural targets.Such macromolecules are, therefore, invalu-able in biotechnology, medicine, and analyticchemistry. However, although “nature’s own,”such structures are far from perfect “tools”—they are unstable out of their native environ-ment and often low in abundance, and a nat-ural receptor for the particular molecule ofinterest may not exist. Researchers have longdreamed of building such structures de novo:creating tailor-made receptors for the desiredmolecular target in bulk. One surprisinglysimple way of generating artificial macromol-ecular receptors is through the molecularimprinting of synthetic polymers. Now, in arecent publication in Nature1, Zimmermannand colleagues have made an exciting newcontribution to this field: they created a mole-cular impression in the core of a dendrimer,obtaining a soluble artificial receptor the sizeof a small protein.

In molecular imprinting, a target molecule(or a derivative thereof) acts as the templatearound which interacting and crosslinkingmonomers are arranged and co-polymerizedto form a cast-like shell (Fig. 1). Initially, themonomers form a complex with the templatethrough covalent or noncovalent interactions.After polymerization and removal of the tem-plate, binding sites complementary to the tar-get molecule in size, shape, and position offunctional groups are exposed and their con-firmation is preserved by the crosslinkedstructure. In essence, a molecular memory isimprinted on the polymer, which is now capa-ble of selectively rebinding the target. Thus,molecularly imprinted polymers (MIPs) pos-sess two of the most important features of bio-logical receptors—the ability to recognize andbind specific target molecules.

Nevertheless, MIPs do differ from biologi-cal receptors: MIPs are large, rigid, and insolu-ble macromolecules, whereas their naturalcounterparts are smaller, flexible, and in most

instances soluble. Depending on their size,MIPs can bear thousands or millions of bind-ing sites per molecule, whereas biologicalreceptors have a few binding sites, or just one.Moreover, the population of binding sites inMIPs, especially those of noncovalentlyimprinted polymers, is heterogeneous becauseof the influence of the equilibria that governthe template–monomer complex formation,and the dynamic of the growing polymerchains before co-polymerization. Althoughnot always problematic, these characteristicscan prevent MIPs from being substituted fornatural receptors in certain applications.

Now, Zimmermann and colleaguesdescribe a system that avoids some of thesepotential shortcomings1. Their strategy is tomake the molecular imprint inside den-drimers—macromolecules of highly regularstructure consisting of a polyfunctional cen-tral core covalently linked to layers of repeat-ing units (so-called generations). In this study,the template molecule and core used was theporphyrine derivative tetrakis-meso(3,5-dihydroxyphenyl)-porphyrine, to which eightthird-generation dendrons were covalentlyattached through ester links. The outer layer ofthe dendrons was composed of homoallyl endgroups, and thanks to these extremities, theouter “shell” of the dendrimer could becrosslinked or “polymerized” intramolecular-ly. Finally, hydrolytic removal of the por-phyrine template liberated the binding sites,which then contained eight precisely posi-

Creating a good impressionBy the imprinting of a molecular memory in their core, dendrimerscan be tailored to bind to defined molecular targets in a selectiveand reversible fashion.

Karsten Haupt

Karsten Haupt is a member of the sciencesfaculty at the University of Paris 12, 94010Créteil, France, and is affiliated with LundUniversity, Sweden(karsten.haupt@tbiokem.lth.se).

Figure 1. Creating a molecular imprint using a synthetic polymer. The interacting monomers,crosslinker, and template molecule are mixed together (left). The interacting monomers assembleinto a complex with the template molecule (here through noncovalent interactions). Polymerization isinitiated, and the interacting monomers co-polymerize with the crosslinker, forming soluble polymerchains. As polymerization proceeds, an insoluble, highly crosslinked polymeric network is formedaround the template. Removal of the template then liberates complementary binding sites. Insetpanel, the different steps of the process are illustrated for the amino-acid derivative Cbz-p-aminophenylalanine as the template and methacrylic acid as the interacting monomer.

©B

ob C

rim

i

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reb

iote

chn

olo

gy

NEWS AND VIEWS

www.nature.com/naturebiotechnology • SEPTEMBER 2002 • VOLUME 20 • nature biotechnology 885

the binding site that, after the template isremoved, allows accommodation of the origi-nal target molecule though noncovalent inter-actions, resulting in a stable imprinting com-plex and fast rebinding of the target.

However, the imprinting of polymers usingcovalent complexes—including this new den-drimer system—involves a substantial amountof time-consuming organic chemistry, and islimited to targets to which monomers can bereadily attached. By comparison, noncovalentimprinting, a concept pioneered by Mosbach4,seems much more straightforward: mixtogether target and suitable monomers, initiatepolymerization, and let the MIP grow itself.This often works, but the noncovalentimprinting approach is not as easy as it sounds.A closer look at the recent literature suggeststhat a substantial amount of work is devoted toimproving noncovalently imprinted polymers.For example, new monomers are beingdesigned that can form stronger noncovalentinteractions with targets. Other researcherspropose to rely on combinatorial methods5 ormolecular modeling6 to find the optimal com-bination of monomers for noncovalentimprinting of a given target. Wulff andcoworkers have recently attempted to obtainMIPs of the same size as proteins, by synthesiz-ing imprinted microgels with molecular mass-es of 104–105 Da (ref. 7).

Industry is currently evaluating the poten-tial application of and commercial opportuni-ties for MIPs, and here, proof of principle isnot the only criteria for future investment.Companies need to investigate the selectivityof MIPs for their targets, and their compatibil-ity with the environment in which they are tobe used, including biological fluids and tissues.Criteria such as the ready integration of molec-ular imprinting within existing industrial fab-rication processes, yields, cost, and the com-petitiveness of MIPs with existing affinitymaterials also need to be examined.

The day when MIPs were only used for theseparation of isomers is long past. AfterMosbach et al.8 described the use of MIPs asantibody mimics in immunoassays, the num-ber of publications in the area rose exponen-tially. Today, the main opportunities for thetechnology are in analytical chemistry9, butinterest is growing in the biomedical field forthe use of MIPs, for example, in drug discoveryor as therapeutics themselves. For example,Mosbach’s group has recently described syn-thesizing new enzyme inhibitors by assemblingand interconnecting different building blocksat the binding sites of a MIP imprinted with aknown inhibitor of the enzyme10.

For many of potential MIP applications, it ishighly desirable, if not essential, for the MIPs tohave well-defined binding sites. Zimmermann

and colleagues’ imprinted dendrimers could beused in many of the traditional applications ofMIPs, either in soluble form, immobilized to asupport, or perhaps tagged with a marker.Several outstanding questions about this tech-nology remain to be answered: is the newmethod flexible enough to be used with differ-ent kinds of templates, including those that arenon-symmetrical? Can they be made water sol-uble? Can stable noncovalent complexes beimprinted in the same manner? Whatever theanswers, one thing is for sure: these imprinteddendrimers are an impressive piece ofsupramolecular chemistry.

1. Zimmerman, S.C., Wendland, M.S., Rakow, N.A.,Zharov, I. & Suslick, K.S. Nature 418, 399–403(2002).

2. Wulff, G. & Sarhan., A. Angew. Chem. 84, 364 (1972).3. Klein, J.U., Whitcombe, M.J., Mulholland, F. & Vulfson,

E.N. Angew. Chem. Int. Ed. Engl. 38, 2057–2060(1999).

4. Arshady, R. & Mosbach., K. Makromol. Chem. 182,687–692 (1981).

5. Takeuchi, T., Fukuma, D. & Matsui, J. Anal. Chem. 71,285–290 (1999).

6. Chianella, I. et al. Anal. Chem. 74, 1288–1293 (2002).7. Biffis, A., Graham, N.B., Siedlaczek, G., Stalberg, S.

& Wulff, G. Macromol. Chem. Phys. 202, 163–171(2001).

8. Vlatakis, G., Andersson, L.I., Müller, R. & Mosbach,K. Nature 361, 645–647. (1993).

9. Haupt, K. Analyst 126, 747–756 (2001).10. Mosbach, K., Yu, Y.H., Andersch, J. & Ye, L. J. Am.

Chem. Soc. 123, 12420–12421 (2001).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reb

iote

chn

olo

gy